BNPL looks like the safest lending product ever invented: tiny tickets, six-week tenor, no interest to the customer, merchant foots the bill. So teams underwrite it like it's barely credit at all — instant yes, minimal checks, scale hard. Then the loss line shows up and it's worse, per dollar lent, than a product with ten times the ticket size. The reason isn't a mystery once you look at the actual economics.

First, how it makes money. BNPL's revenue engine is the merchant discount rate — the merchant pays a few percent because BNPL lifts conversion and basket size. Late fees and, on longer plans, interest are secondary. The point for credit people: revenue per loan is small, and the tenor is short. That single fact shapes everything about the risk.

Here's where the loss actually comes from.

1. Thin margins mean no cushion

A few percent of MDR on a six-week loan is a sliver of revenue. A personal loan earns interest over years; BNPL earns once, briefly. So a one-point rise in default rate that a personal-loan book would shrug off can erase BNPL's entire margin. The economics are unforgiving by design — there's almost no room for credit error.

2. Losses are front-loaded — first-payment default and fraud

Instant approval and low friction mean a large share of losses are first-payment defaults and fraud, not seasoned credit deterioration. You lose early — before you've earned anything on the customer. The "soft" instant yes is the loss lever.

3. Invisible stacking

Customers run multiple BNPL plans across providers, and historically thin bureau reporting means you can't see their total BNPL leverage. You approve someone already over-extended at five other providers. That's structural, not bad luck — and it's the failure mode regulators are now zeroing in on.

4. Cohort drift as you scale

Your first cohorts look pristine because early adopters self-select. Push for growth — broaden the base, raise purchase frequency — and loss rates climb, often right after the board has extrapolated the lovely early numbers into the plan.

5. Cost-to-collect on tiny balances

Chasing a $40 missed installment costs nearly what chasing a $4,000 one does. Operational cost eats the already-thin margin. "Low balance" does not mean "low cost."

6. Regulation compressing the model

Late-fee caps and new affordability rules — the FCA's BNPL regime lands in 2026, with others following — cut fee revenue and raise underwriting cost at the same time. The light-touch era is ending, and the model has to absorb it.

Most BNPL teams underwrite the customer and stop there. But loss and fraud cluster hard by merchant and category — the same applicant is a very different risk buying electronics from a flash-sale site than buying groceries from an established retailer. Managing risk at the merchant level — pricing, limits, and monitoring by merchant category — is the underused lever. Pair it with dynamic customer limits that expand only for proven repayers and choke off stackers fast, and you plug the two biggest leaks at once.

Stop staring at a single blended loss rate. Cut your losses by merchant category and by cohort vintage. That one view almost always reveals that a thin slice of merchants and your newest, broadest cohorts are carrying most of the loss — and that's something you can act on next week.

Next week: building a scorecard with alternate & cash-flow data — scoring the customers a traditional bureau can't see.

If this was useful, forward it to someone pricing a BNPL book.

Views are my own and don't represent my employer.